Information processing systems incorporating virtualization infrastructure are coming into increasingly widespread use. An issue that arises in such systems relates to network virtualization. For example, in a typical network virtualization arrangement, distributed compute nodes of a given system can be implemented using geographically dispersed portions of different processing platforms supporting the virtualization infrastructure. The topology of the compute nodes can change frequently due to variations in computational demands within the system. However, movement of compute nodes within such a system can be problematic in that it can require unduly complex network address translation operations. In addition, use of conventional network address translation can in some cases prevent the implementation of location-based service functionality for the compute nodes.